Comparative diagnostic capability of large language models in neurosurgery
ChatGPT-4 offered the most accurate diagnoses when given established clinical vignettes. Adding imaging processing capabilities and relevant data significantly increased the accuracy of LLM diagnoses. LLMs can offer accurate assessments of common neurosurgical conditions but necessitate detailed prompting from clinicians. Artificial intelligence has incredible clinical potential; however, practitioners must be cautious and think critically while using them for diagnostic purposes.
Authors
- Akshay Warrier (ORCID: https://orcid.org/0000-0003-4841-6366)
- Caryn J Ha (ORCID: https://orcid.org/0000-0002-8867-1217)
- Shrey Shah
- Yaxel Levin-Carrion
- Rohan Singh (ORCID: https://orcid.org/0000-0001-6532-1639)
- James K. Liu (ORCID: https://orcid.org/0000-0002-5952-4082)
- Amanda Bosland
- Joseph Wardell
- Tannishtha Som
- Alejandro Pando
Institutions
- Rutgers, The State University of New Jersey (US)
- Saint Barnabas Medical Center (US)
- North American Skull Base Society (US)
Publication Details
- Journal
- Journal of neurosurgery
- Published
- 2026-02-01
- DOI
- https://doi.org/10.3171/2025.9.jns25846
- Primary Topic
- Artificial Intelligence in Healthcare and Education
- Type
- article
- Field-Weighted Citation Impact
- 0.00